Forum

Author Topic: Difference True Ortho vs Normal Ortho - Assign images tool  (Read 3189 times)

obw

  • Newbie
  • *
  • Posts: 45
    • View Profile
Difference True Ortho vs Normal Ortho - Assign images tool
« on: August 16, 2017, 02:13:16 PM »
Hi there.

I found an old topic regarding this matter (http://www.agisoft.com/forum/index.php?topic=3850.0) but it didn't give me the answers I was looking for.

Is the Ortomosaics created from Agisoft a True Ortophoto that can be used to measurement purposes?

My model has markers and is georeferenced (NTM10), but when the Ortho is created the buildings does not look good, see attached picture. The distances on the roof edges on the same building can vary up to 1 meter. The ground all looks good. The flight was flown with 75 % overlap in Dronedeploy, while the Agisoft report says a average overlap of between 4,67 - 7,3035.

I've played around with the Assign images tool and it gives me the opportunity to change the image and correct buildings so it all LOOKS Nice, but I have no idea if the measurments are good? I don't know which pictures to choose since some of them are showing the facades of bulidings. If I use this tool - Is the image no longer a "True Orthophoto"?

How do you get buildings to look better in the automated process?

Thanks for all your help, looking forward to hear from you!
« Last Edit: August 16, 2017, 02:18:46 PM by obw »

SAV

  • Hero Member
  • *****
  • Posts: 710
    • View Profile
Re: Difference True Ortho vs Normal Ortho - Assign images tool
« Reply #1 on: August 16, 2017, 04:17:17 PM »
Hi obw,

The overlap of 75% that is chosen in Dronedeploy is based on a flat surface (from where you take off). In your case there are buildings 'sticking out' of the ground which will lead to a reduced image overlap in these elevated areas. This is one of the reasons why your roof areas are not properly reconstructed. You need to increase your image overlap (forward and side overlap). Additionally, you might want to change your survey layout/design. Fly a grid pattern and capture slightly oblique imagery.

This should solve your issues.

Regards
SAV